From owner-freebsd-stable  Sat Feb 24 16:30: 7 2001
Delivered-To: freebsd-stable@freebsd.org
Received: from resentment.verloid.net (90.ppp1-8.osl1.world-online.no [213.142.76.218])
	by hub.freebsd.org (Postfix) with ESMTP id 2803D37B65D
	for <freebsd-stable@freebsd.org>; Sat, 24 Feb 2001 16:30:03 -0800 (PST)
	(envelope-from needle-mls@world-online.no)
Received: from resentment.verloid.net (localhost.verloid.net [127.0.0.1])
	by resentment.verloid.net (8.9.3/8.9.3) with SMTP id BAA01760
	for <freebsd-stable@freebsd.org>; Sun, 25 Feb 2001 01:29:56 +0100 (CET)
	(envelope-from needle-mls@world-online.no)
From: needle-mls@world-online.no
Received: by resentment.verloid.net (sSMTP sendmail emulation); Sun, 25 Feb 2001 01:29:56 +0100
Date: Sun, 25 Feb 2001 01:29:56 +0100
To: freebsd-stable@freebsd.org
Subject: libcrypto hosed when compiled with optimizations
Message-ID: <20010225012956.A1726@resentment.verloid.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
X-Operating-System: FreeBSD 3.5-STABLE
Sender: owner-freebsd-stable@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Hi,

On my 4.2-stable system I have tweaked my /etc/make.conf to compile
sources with -O -mcpu=686 -march=pentiumpro -funroll-loops and as
far as I can tell, everything is working flawlessly except for ssh
with protocol v2. For some reason when I ssh -2 localhost, or ssh -2
anywhere else I get the following error:

% ssh -2 -v arrakis
(...)
debug: Host 'arrakis' is known and matches the DSA host key.
debug: bits set: 499/1024
debug: len 55 datafellows 0
debug: dsa_verify: signature incorrect
dsa_verify failed for server_host_key
debug: Calling cleanup 0x80592a0(0x0)

Ssh'ing from a computer with unoptimized binaries to my
workstation, using protocol v2, works so sshd is obviously fine.
Recompiling libcrypto without -O -mcpu etc. fixes the problem.
It's obviously a little picky about how it's compiled.

Now I'm wondering if my foot wound is more serious, could anything
else have blown up as well or is the crypto code just more touchy?
I use the same optimizations on the kernel, while ports are compiled
with -O2 and as I mentioned earlier I have not noticed anything else
not working.
Has anyone done any testing which shows that things actually run
faster, or are the binaries just more likely to be flawed?

For the record, this is 4.2-stable as of Feb. 22 with OpenSSH 2.3.0.

Regards,

Jo

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message